This paper presents a novel system integration technical risk assessment model (SITRAM), which is based on Bayesian belief networks (BBN) coupled with parametric models (PM). This model provides statistical information for decision makers, improving risk management of complex projects. System integration technical risks (SITR) represent a significant part of project risks associated with the development of large software intensive systems for defense and commercial applications. We propose a conceptual modeling framework to address the problem of SITR assessment in the early stages of a system life cycle. Initial risks' taxonomy and risks' interrelations have been identified using a hierarchical holographic modeling (HHM) approach. The framework includes a set of BBN models, representing relations between risk contributing factors, and complementing PMs, used to provide input data to the BBN models. In this paper, we present the rationale and the modeling objectives, and describe the concepts and details of BBN experimental model design and implementation. To address practical limitations, heuristic techniques have been proposed for easing the generation of conditional probability tables. PM design principles are described and examples are presented. In conclusion, we summarize the benefits and constraints of SITR assessment based on BBN models. Further research directions and model improvements are also presented.
|Number of pages
|IEEE Transactions on Systems, Man and Cybernetics: Systems
|Published - Mar 2014